Richard Abrich
Richard Abrich
### Feature request We want to support ollama, since they provide local multimodal inference: https://github.com/jmorganca/ollama/releases/tag/v0.1.15 ### Motivation Vision is most general.
### Describe the bug When running `python -m openadapt.visualize`: ### To Reproduce set `config.VISUALIZE_RUN_NATIVELY = False`, then: ``` poetry shell python -m openadapt.visualize ``` Note that eventually the visualization opens...
### Feature request How can we incorporate https://github.com/outlines-dev/outlines ? This likely involves replacing the `CompletionProvider` API. ### Motivation Outlines 〰 is a library for neural text generation. You can think...
### Feature request We would like to migrate from Gitbook / Google Docs to MkDocs. ### Motivation Gitbook requires a subscription, while https://squidfunk.github.io/mkdocs-material/ is completely free.
### Feature request We would like to implement Instructor (https://jxnl.github.io/instructor/) to automate the creation of objects from prompts ### Motivation _No response_
### Feature request How can we implement: > dynamic few-shot selection, self-generated chain of thought, and choice shuffle ensembling From https://arxiv.org/abs/2311.16452 See also https://twitter.com/erichorvitz/status/1729854235443884385 ### Motivation > We introduce Medprompt,...
### Role Clinical Psychiatrist ### Task How can we enable Clinical Psychiatrists to automate repetitive tasks? This involves: - Identifying specific use cases relevant to clinical psychiatry (e.g. clinical documentation,...
### Feature request How can we incorporate ideas from https://github.com/PKU-YuanGroup/Video-LLaVA ? ### Motivation > With the binding of unified visual representations to the language feature space, we enable an LLM...
How can we enable loss adjusters (and their managers) to automate repetitive tasks? This task involves creating recordings involving one or more relevant applications, and implementing tests for corresponding completions:...
### Feature request How can we incorporate the ideas in https://craftjarvis-jarvis1.github.io/ ? ### Motivation > We introduce JARVIS-1, an open-world agent that can perceive multimodal input (visual observations and human...